Strategic Risk and Co-Operative Relationships: Experimental Evidence
23 Pages Posted: 20 Oct 2008
Date Written: October 17, 2008
We test infinitely repeated prisoner's dilemma games with random continuation in the laboratory to capture the effect of strategic risk on co-operation. We propose a criterion building on Harsanyi and Selten's (1988) risk dominance concept and motivate it by three heuristic principles. Our criterion depends on the often disregarded "sucker's payoff" - the payoff of a co-operating player meeting a defector - and results in a critical discount factor delta* strictly larger than delta, the traditional criterion, above which co-operation is supported by Pareto-undominated equilibrium strategies and which has often been used in applications.
We find that changes in the frequency of subjects' co-operation in the lab are predicted far better by delta* than by the Pareto-dominance criterion reflected in delta. Our experiments, in particular, show that for parameter changes where the two criteria predict co-operation frequency changes in opposite directions, our alternative criterion delta* is the one that predicts correctly.
Keywords: co-operation, repeated Prisoner's Dilemma, experiments, strategic risk, risk dominance, sucker's payoff, collusion, corruption
JEL Classification: C72, C73, C92
Suggested Citation: Suggested Citation